Databases
StackOverflow2013
Postsdbo
PO
 

Note that there are some explanatory texts on larger screens.

plurals
  1. PO
    primarykey
    data
    text
    <p>If you have access to atomic reads and atomic writes on your platform(s) then create a work allocation table (read the architecture notes for your platforms - it may or may not be that ordinary reads and writes are good enough, you may or may not need to add memory barriers):</p> <p>One byte per line, initially zero, non-zero means the thread the line is allocated to</p> <p>...and create an atomically updated count of lines done int field per worker thread. The table should be updated and read using atomic read/write instructions (so in chunks of 8,16,32 or 64bits depending on the available instructions on the platform).</p> <p>The top level logic must work out whether to just do all the work straight away on the main thread (if the image is really small) or to start one worker thread, or to start N worker threads.</p> <p>The coordination thread (or if it was me probably I'd scrap the coordination thread and do this on the main thread) allocates half the jobs in round robin to the threads (or all the work if there's less than a certain amount). If it allocates less than all the work it monitors the other threads and benchmarks the performance of average thread and best thread. It makes sure that threads as a whole do not run out of work, but tries not to leave threads with nothing to do.</p> <p>The front-end keeps a pointer for each worker to where in the allocation table a worker has done up to, and when the worker increases its integer field for how many lines it has done, the front-end searches forwards through the work allocation table finding the line indices of the jobs allocated to that worker that are now completed, and updates a bit buffer of which specific lines are done and also updates a total done field.</p> <p>--</p> <p>This is a general algorithm for dynamically allocating the work to the threads, as another poster has suggested you can alternatively statically allocate the work by making the line numbers the worker thread should process a function of the number of worker threads and the worker thread number, then just pass count of lines done by each worker through an atomic field to the front end.</p>
    singulars
    1. This table or related slice is empty.
    1. POHow can my threaded image generating app get it's data to the gui?
      singulars
      1. PTQuestion
    1. PTAnswer
    1. This table or related slice is empty.
    1. USmartinr
    plurals
    1. This table or related slice is empty.
    1. This table or related slice is empty.
    1. POHow can my threaded image generating app get it's data to the gui?
      singulars
      1. PTQuestion
    1. This table or related slice is empty.
    1. VO
      singulars
      1. PO
      1. This table or related slice is empty.
      1. VTAcceptedByOriginator
    1. COThanks martinr for the effort you put into these answers. Although I don't understand them as much as I hoped, they've given me some ideas to try out.
      singulars
      1. PO
      1. USJames Morris
    2. COIt's unlikely my implementation will make any decisions about the amount of work to do, especially based on image size. While the image size does factor into rendering time, the minuteness of the area being rendered, the number of iterations (per pixel) to be calculated, and the overall detail complexity of that area also greatly affect how much time is spent rendering, such that even for a small 200 x 180 image, rendering time could be improved by using more than one thread.
      singulars
      1. PO
      1. USJames Morris
 

Querying!

 
Guidance
A row detail

Detail views are divided into sections. All the information in the data section comes from columns in the selected row. The other sections display data from other, related rows.

Related data can be related in a to-one or a to-many fashion. Captions of data related in a to-many fashion link to a list view showing a filtered view of the table.

Try moving around until you find a non-empty to-many entry and click on the label to get to one. You can move back to the root by clicking on the database name in the header.

SQuiL has stopped working due to an internal error.

If you are curious you may find further information in the browser console, which is accessible through the devtools (F12).

Reload